International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 4 Issue 1, December 2019 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
Adaptive Question Recommendation
System Based on Student Achievement
Çakir, A.1; Yilmaz İnce, E.2; Babanyyazova, Y.1
1Department
of Electric-Electronic Engineering, 2Department of Computer Technologies,
1,2Isparta University of Applied Sciences, Isparta, Turkey
How to cite this paper: Çakir, A. | Yilmaz
Ince, E. | Babanyyazova, Y. "Adaptive
Question Recommendation System Based
on Student Achievement" Published in
International Journal
of Trend in Scientific
Research
and
Development (ijtsrd),
ISSN:
2456-6470,
Volume-4 | Issue-1,
December
2019,
IJTSRD29789
pp.1034-1038, URL:
www.ijtsrd.com/papers/ijtsrd29789.pdf
ABSTRACT
In this research, it is aimed to determine the success level of the students
according to the answers given in the previous exams with TOPSIS method, to
send the question to the related student by the authorized teacher according
to the determined success level, to analyze the student's periodical (weeklymonthly-yearly) performance with graphical data and to create a data bank
and to determine the overall performance of the student. Developmental
research technique was used, which is one of the design-based research
method derivatives including the design of educational tools. This study
consists of question suggestion and question selection tool components.
Adaptive question suggestion system perform generation, storage,
classification, disclosure and publication of the content. The software has been
successfully tested by 350 students and teachers by using 1000 questions
prepared by the educator about the courses and subjects previously defined.
As a result, it is seen that students check all the questions related to courses
and subjects in the application database and submit the most appropriate
question to the requestor in response to the question requests. Also it is
observed that the TOPSIS algorithm quickly determined the desired number of
questions from the question bank formed by the authorized teacher in
accordance with the required criteria, and the process of creating a trial or
subject test was performed successfully.
Copyright © 2019 by author(s) and
International Journal of Trend in Scientific
Research and Development Journal. This
is an Open Access article distributed
under the terms of
the
Creative
Commons Attribution
License
(CC
BY
4.0)
(http://creativecommons.org/licenses/by
/4.0)
KEYWORDS: adaptive question recommendation system; student achievement;
multi-criteria decision making technique; developmental research; topsis
1. INRODUCTION
Today, technology development and diversity related to
education of technology is increasing. The most up-to-date
topics are that students can access e-learning content not
only at school but also at other times and participate in
constructivist education that will enable the active
participation of students (Kong et al., 2014) Adaptive
Learning Process is a computer based education method that
organizes and plans the education of each student in the
most appropriate way considering various needs situations
(Szijarto and Cousins, 2019).The question recommendation
system is the process of evaluating the system for a purpose
by means of automatic or manual tuning and trial process,
and observing the number of answers given as expected
because it meets the specified response rate, and presenting
the results to the system by determining the differences
between them (Hansen et al., 2019). In this study, a system
has been developed including the functions of storing,
classifying, modifying, searching and listing the questions
using TOPSIS algorithm, creating, modifying and publishing
new contents using the questions prepared again.
2. MATERIALS AND METHODS
In this paper, developmental research technique was used,
which is one of the design-based research method
derivatives including the design of educational tools. This
study consists of question suggestion and question selection
tool components. Adaptive question recommendation
system perform generation, storage, classification, disclosure
@ IJTSRD
|
Unique Paper ID – IJTSRD29789
|
and publication of the content. The main features of the
question recommendation system are presented in Table 1.
The developed software sends learning questions to
students, answers them, and provides a level of achievement.
Learning questions The most convenient and accurate way
to mark the answers to test questions in SQL repositories
can be provided by data standards. For this reason, some
standardization studies were done on the data of learning
questions (Stratakis et al., 2003).
Table1. The main features of the question
recommendation system
Feature
Description
Search or
Be able to easily find Learning Tests
Find
with the data entered
The Question Repository should
Quality
provide a technical and pedagogical
control
requirement comparison of Data and
Test questions entered into the System
Presentation Test questions should be able to be sent
process
to students' smart devices
Provide comprehensive search by
Addition
gathering information from other
question pools
Question Repositories Must contain
Request
different options to provide test
questions for learning
Volume – 4 | Issue – 1
|
November-December 2019
Page 1034
International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
The developed software was developed with the aim of
storing, searching, sharing, and selecting the appropriate test
question for learning purposes. To accomplish these
functions, an MSSQL Server 2016 database management
system and LINQ to SQL and DataSet.xsd techniques can be
used in conjunction with C # and MSSQL. These features are
prepared with ASP.Net technology. This software, which
processes according to the student achievement, allows the
addition of learning objects for any different academic
studies.
3. ADAPTIVE QUESTION RECOMMENDATION SYSTEM
Instructors use the Adaptive Question Recommendation
System (AQRS) software to automatically send test question
contents to the system of the course students, to make
assessments about their study notes or students, and to
access this recorded information at any time. In the software,
relational database applications, WebApi applications and
Another purpose of the software is to store learning objects
and tests in the database and use them when needed.
There are three types of user roles in AQRS software that are
administrator, instructor and student. In order to add any
objects or use other features of the software, users must
register to the system. Membership requests are sent
through the user of the Windows Form base. Dev Express
(v18.2.6) The Win Form library functions to interface the
software, and then administrators examine these requests to
allow appropriate user roles. After the administrator has
entered the usage role by the administrator, users can use
the AQRS software. Learning tests and objects are stored in
the system via MSSQL Server interfaces. If necessary,
learning can be easily accessed by testing and listing method
by data or types of objects.
In the AQRS software, the necessary components for the
learning test and object data were prepared. In addition, any
data format can be stored in the system as a JSON-based file
format and can be converted to Web Api format at any time
and sent. Users can answer, analyze, comment on learning
objects and tests, and identify problems using their
interfaces. One of the most important features of the AQRS
software is that the instructor approves the quality control
of test questions and objects in the system. After checking
the test questions prepared for learning purposes, the test
questions have the feature that all students registered in the
software can be transmitted to their mobile devices. The
main page of the authorized instructor and the unauthorized
instructor for access to the system shown in Figure 1 is as
follows.
@ IJTSRD
|
Unique Paper ID – IJTSRD29789
|
Figure1. Instructor and the unauthorized instructor
for access to the system
3.1. Architecture of AQRS software
The developed AQRS software is based on a multi-layered
software architecture and an object-oriented programming
approach. There are two horizontal layers in the upper layer
architecture of the software. These are both the presentation
layer and the data layer. These layers have the component
structures required for the system operational state of the
AQRS software shown in Figure 2. The existence of these two
layers makes the software project modular. For example,
extra component structures can be added to the software
simply and quickly. The connection between the database
and the application in the data layer is prepared.
The Entity Framework sets an open source object option for
the ADO.NET and C # .NET programming language. In Figure
2, instructors can use these interfaces to add, list, and view
the data of learning tests using the Windows Form base. In
the work layer, there are operations of the data conversion
engine. When a request is received by the presentation layer,
the system makes a decision and transmits it to the data
layer in order to process the workflow state that appears in
the software faster. The data layer adds comments about the
instructor and students, reporting problems, users and roles,
and all data such as the topic and the sub-subject that fits the
database.
Volume – 4 | Issue – 1
|
November-December 2019
Page 1035
International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
TOPSIS algorithm is used in AQRS software according to
student achievement level. This decision-making criterion is
based on the theory of priority. Also, problems involving
complex decision-making criteria can be easily solved with
TOPSIS. It is useful for logically using the data entered into
the system and the decisions of experts. Since the TOPSIS
algorithm does not contain mixed mathematical expressions,
it is easy to use and has a transparent structure for decision
makers because of the sorting criteria. The steps for
selecting adaptive questions with TOPSIS are shown as
follows.
In the TOPSIS method, the state of formation of the decision
matrix (A), the state of formation of the standard decision
matrix (R), the state of formation of the weighted standard
decision matrix (V), the states of formation of the solutions
in the ideal direction
and the negative ideal direction
, It consists of a number of steps, such as calculating the
degree of discrimination criteria and calculating the relative
proximity to the ideal solution.
Figure2. Architectural structure of AQRS
3.2. Selection of Tests with TOPSIS Algorithm
Trainers and students try to find the most appropriate
learning objects from many sources (Sinclair et al., 2013).
For this reason, it is necessary to carry out efficient studies
on the quality of tests for learning purposes. The quality of
learning objects or tests is difficult to measure because
instructors and learners can respond according to the
requirements and requirements to which they depend.
There are many selection methods to find the test that is
most suitable for learning. There are many types of software
to measure the quality status of tests for learning purposes
(Vargo et al., 2003).
The AQRS application is set to submit test questions by
evaluating the status of measurement and analysis according
to the achievement status, interaction, participation,
availability and content categories of the students with
TOPSIS algorithm. According to the scoring results, there are
various softwares that use multi-criteria evaluation methods
to select the best alternative test for learning (Kurilovas,
2009). There are cases where each multi-criteria decision
making technique is superior to other criterion techniques
(Ömürbek et al., 2013). For this reason, which multi-criteria
decision-making method should be applied to the decision
problem, natural structure of decision problem, various
options, scales of evaluations, type status of uncertain rate,
dependency ratio of criteria, expectation status of the
decision maker, data quantity and quality status to be
entered factors should also be taken into consideration
(Ceiling and Hatami-Marbini, 2011).
TOPSIS algorithm is applied in production, marketing,
planning, group decision making, distribution of resources,
education, health, selection of markets, transportation,
support of decisions and many other areas (Geuna and
Martin,2003). The TOPSIS algorithm is known as one of the
most excellent multi-criteria decision making methods
because it has the advantage of being able to sort out the the
excessively large number of alternatives and to quickly
determine the best alternative outcome (Tavana and
Hatami-Marbini, 2011).
@ IJTSRD
|
Unique Paper ID – IJTSRD29789
|
Step 1: First, the decision matrix (A) needs to be established.
The evaluation matrix is started by creating the decision
matrix in which alternative values are required to be listed
in the rows and criteria to be evaluated based on the
decision-making evaluation in the columns. The step by the
decision makers, known as the initial matrix, which consists
of alternative and n criterion evaluations, is as follows:
Step 2: The standard decision matrix (R) needs to be
established. The initial values that take place in our decision
matrix must be brought to the standard state.
Step 3: Since the weighted standard decision matrix (V)
of the
needs to be established, the weighting ratios
criterion assessments must first be determined.
Then, each individual value found in the R matrix must be
multiplied by the values of the criterion
to which it is
connected and a Weighted Standard Decision Matrix (V)
should be created.
and
Step 4: Solution results in the ideal direction
negative in the ideal direction
should be established.
The TOPSIS method assumes that each criterion assessment
has shown a tendency to increase or decrease steadily. For
this reason, the maximum value in each column in the matrix
V represents the ideal direction
solution, the minimum
value is the negative ideal solution
solution.
Volume – 4 | Issue – 1
|
November-December 2019
Page 1036
International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
Step 5: The state of calculation of the separation criteria
degrees. The distance in the Euclidian Approach is used to
calculate the distance of the criterion at each decision point
to the solution point in the ideal direction and the negative
ideal direction. According to these values, the distance of the
ideal solution in the ideal solution and the distance of the
ideal solution in the negative direction are calculated by
the formulas shown below.
Step 6: Calculate the degree of proximity relative to the ideal
solution result. Using the distinction criteria in the ideal
direction and in the negative ideal direction, the relative
proximity of each alternative case to the ideal solution result
is calculated.
i= 1, …, m
The determination of the criteria for selecting the test
question for learning purposes and other criteria belong to
the administrator or trainer and the criteria are adjusted
according to the test statistics. These criteria can be
evaluated manually by instructors by manual input or by
decision-making methods.
The trainers were interviewed to determine the selection
criteria of the data-based test questions. Ten instructors
participated in the interview. According to the data in the
interview, TOPSIS question coefficient criteria (reliability,
discrimination, cognitive area, degree of difficulty of the
question) were determined for the selection of test
questions.
Test questions with the highest value obtained by TOPSIS
algorithm are selected. As shown in Table 2, the AQRS
software was tested with four criteria. In the first few trials,
the CRs of the criteria were found to be greater than 0.2,
indicating that the criteria did not proceed consistently with
the selection of test questions. At the end of the fifth trial, the
CR of the criteria was found to be 0.877802262, which was
below 0.877, and the criteria and points given were
therefore consistent. When necessary, instructors can
change the number of criteria up to 10 or even more and
enter alternatives according to preference. Test questions
are evaluated according to four different criteria.
Table2. Criteria weights
Criteria weights
Question
Number
C1
C2
C3
C4
0,1
0,3
0,4
0,2
Q14
6
3
5
Q8
5
3
Q9
5
Q13
4
Total
S-
S+
4
0,877802262
0,012309
0,088422
4
3
0,719027294
0,028831
0,073780
3
4
3
0,719027294
0,028831
0,073780
2
5
4
0,711699469
0,030011
0,074085
4. CONCLUSION
In this research, it is aimed to determine the success level of
the students according to the answers given in the previous
exams with TOPSIS method, to send the question to the
related student by the authorized teacher according to the
determined success level, to analyze the student's periodical
(weekly-monthly-yearly) performance with graphical data
and to create a data bank and to determine the overall
performance of the student. The software has been
successfully tested by 350 students and teachers by using
1000 questions prepared by the educator about the courses
and subjects previously defined. As a result, it is seen that
students check all the questions related to courses and
subjects in the application database and submit the most
appropriate question to the requestor in response to the
question requests. Also it is observed that the TOPSIS
algorithm quickly determined the desired number of
questions from the question bank formed by the authorized
teacher in accordance with the required criteria, and the
process of creating a trial or subject test was performed
successfully.
@ IJTSRD
|
Unique Paper ID – IJTSRD29789
|
REFERENCES
[1] Geuna, A., Martin, B. R. University Research Evaluation
and Funding: An International Comparison. Minerva,
41(4): 277-304, (2003).
[2] Hansen, P., Bustamante, R. J., Yang, T. Y., Tenorio, E.,
Brinton, C., Chiang, M., & Lan, A. (2019, July). Predicting
the Timing and Quality of Responses in Online
Discussion Forums. In 2019 IEEE 39th International
Conference on Distributed Computing Systems
(ICDCS) (pp. 1931-1940). IEEE.
[3] Kong, S. C., Chan, T. W., Griffin, P., Hoppe, U., Huang, R.,
Kinshuk, Looi, C. K., Milrad, M., Norris, C., Nussbaum,
M., Sharples, M., So, W. M. W., Soloway, E., Yu, S. (2014).
E-learning in School Education in the coming 10 Years
for Developing 21st Century Skills: Critical Research
Issues and Policy Implications. Educational Technology
& Society, 17 (1), 70–78.
[4] Kurilovas, E.: ‘Evaluation and optimisation of elearning software packages: learning object
repositories’. Proc. Fourth Int. Conf. on Software
Volume – 4 | Issue – 1
|
November-December 2019
Page 1037
International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
Engineering Advances, Porto, Portugal, September
2009, pp. 477–483.
[5] Sinclair, J., Joy, M., Yau, Y. J., Hagan, S.: ‘A practiceoriented review of learning objects’, IEEE Trans. Learn.
Technol., 2013, 6, (2), pp. 177–192
[6] Stratakis, M., Christophides, V., Keenoy, K.,
Magkanaraki, A.: ‘E-learning standards, SeLeNe’, ELearning Standards, Version 1, 2003
[7] Szijarto, B., & Cousins, J. B. (2019). Making Space for
Adaptive
Learning. American
Journal
of
Evaluation, 40(2), 160-176.
@ IJTSRD
|
Unique Paper ID – IJTSRD29789
|
[8] Tavana, M., Hatami-Marbini, A. A group AHP-TOPSIS
Frame Work for Human Space Flight Mission Planning
at NASA. Expert Systems with Applications, 38(11):
13588-13603, (2011).
[9] Ömürbek, N., Demirci, N., Akalin, P. Analitik Ağ Süreci
ve TOPSIS Yöntemleri ile Bilimdalı Seçimi. Kilis
Üniversitesi, İİBF, Akademik Araştırmalar ve
Çalışmalar Dergisi, 5(9): 118-140, (2013).
[10] Vargo, J., Nesbit, J. C., Belfer, K., Archambault, A.:
‘Learning object evaluation: computer mediated
collaboration and inter-rater reliability’, Int. J. Comput.
Appl., 2003, 25, (3), pp. 1–8.
Volume – 4 | Issue – 1
|
November-December 2019
Page 1038